ClearType for Squeak/Sub-pixel antialiasing
Tim Rowledge
rowledge at interval.com
Thu Oct 7 16:34:47 UTC 1999
On Thu 07 Oct, Henrik Gedenryd wrote:
> Tell me how this works. It seems to be a different technique than
> *sub-pixel* rendering, and I assume you don't mean regular (gray level, or
> rather intensity) anti-aliasing (isn't 'AA' a strange name btw). (Check out
> that web page. How is it different from regular AA?
My understanding of the term 'sub-pixel anti-aliasing' is that is
effectively a sub-sampling technique that would render the glyphs in a
larger size than you really want and then sample down with
anti-aliasing. It's certainly how we used the term when I was with IBM
research many (many) years ago but it is possible it has been redefined
since then.
So to do vertical sub-pixel a-a, you would render the glyphs several
times over, once for each suitable interval of a pixel fraction times
the scale factor used. If one were using a scaling factor of 4, then you
would render the glyphs 4 times over, once for each of the 4 possible
positionings. Then you'd have to do the same for each of the 4
horizontal positions as well. I have no doubt that you can actually do
it much more cleverly than really doing all this work!
> I also just found out that hinting and grid-fitting in TT is done by a
> 200-opcode hinting language requiring its own interpreter, ouch!
Cute!
tim
--
Strange OpCodes: IOP: Insult OPerator
Tim Rowledge: rowledge at interval.com (w) +1 (650) 842-6110 (w)
tim at sumeru.stanford.edu (h) <http://sumeru.stanford.edu/tim>
More information about the Squeak-dev
mailing list
|